Archimedeans

[Santymax98]

Hello, I think I fixed the equations and the calculation of the tau corresponding to these copulas.

[codecov]

Codecov Report

Attention: 21 lines in your changes are missing coverage. Please review.

Comparison is base (2dfce3d) 79.64% compared to head (a3ca09a) 80.40%.
Report is 3 commits behind head on main.

Files	Patch %	Lines
src/ArchimedeanCopulas/AMHCopula.jl	54.16%	11 Missing ⚠️
src/ArchimedeanCopulas/InvGaussianCopula.jl	74.07%	7 Missing ⚠️
src/ArchimedeanCopulas/FrankCopula.jl	85.71%	2 Missing ⚠️
src/ArchimedeanCopulas/GumbelBarnettCopula.jl	90.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #73      +/-   ##
==========================================
+ Coverage   79.64%   80.40%   +0.75%     
==========================================
  Files          27       27              
  Lines         570      643      +73     
==========================================
+ Hits          454      517      +63     
- Misses        116      126      +10     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lrnv]

Looks like the gumbelBarnett tau does not return anything ? the result line is commented

[lrnv]

Thanks. Could you add a few coherence tests on these two functions ? e.g. test that tau(tauinv(x)) == x for two or three well-chosen x values.

[Santymax98]

I'm going to work with that El El vie, 17 nov 2023 a la(s) 12:30, Oskar Laverny < ***@***.***> escribió:

…

Thanks. Could you add a few coherence tests on these two functions ? e.g. test that tau(tauinv(x)) == x for two or three well-chosen x values. — Reply to this email directly, view it on GitHub <#73 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A7PTWNTOGR2ZLWJ4PPC2CKLYE57KRAVCNFSM6AAAAAA7P6IUQ2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJWGYZTQNJXGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[Santymax98]

Hello @lrnv , I was trying to use the tests, however I am having a problem on my machine with the quadgk function telling me that it is not defined.

[lrnv]

Yes you need to add QuadGK to the dependencies of the package ! While in the package repositoryin julia, use ``] add QuadGKand it will modify theProject.toml` file, that you can then commit. I will do it for you if you want but RN i'm on the bus back home, maybe tomorow

[lrnv]

I pushed to your branch with the fix for the missing package. But I still have weird behaviors in the tests... Cheking by hand,

julia> Copulas.τ(GumbelBarnettCopula(2,0.7))
-0.2726863504658321

whihc is really weird since it is supposed to be only positive dependence for this copula ?

-> The integrated function might be wrong ?

Same thing for the InvGaussian.

[Santymax98]

This copula represents negative dependence. can you see, joe 2014 page 435 El El vie, 17 nov 2023 a la(s) 15:58, Oskar Laverny < ***@***.***> escribió:

…

***@***.**** commented on this pull request. I pushed to your branch with the fix for the missing package. But I still have weird behaviors in the tests... Cheking by hand, julia> Copulas.τ(GumbelBarnettCopula(2,0.7)) -0.2726863504658321 whihc is really weird since it is supposed to be only positive dependence for this copula ? -> The integrated function might be wrong ? Same thing for the InvGaussian. — Reply to this email directly, view it on GitHub <#73 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A7PTWNQSGNOVFJNMR5754XLYE6XVRAVCNFSM6AAAAAA7P6IUQ2VHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMYTOMZXGY4TCOBWG4> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[lrnv]

So I marked the test as Broken... this makes it pass, but this is not good, something must be done about those tests to pass them into "non-broken" state again, this is not acceptable.

[lrnv]

Indeed gumbelbarnett allows negative dependency, i corrected that sorry. But the tests are still broken :)

[Santymax98]

I understand, what can we do? The copula is to model negative dependence, unlike the Clayton copula, where negative dependence is not useful for statistical modeling (since its density is zero in those cases). The Gumbel Barnett copula is very important to deal with that type of dependence. El El vie, 17 nov 2023 a la(s) 16:21, Oskar Laverny < ***@***.***> escribió:

…

Indeed gumbelbarnett allows negative dependency, i corrected that sorry. But the tests are still broken :) — Reply to this email directly, view it on GitHub <#73 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A7PTWNRIKROVZHQMTTOSCILYE62NNAVCNFSM6AAAAAA7P6IUQ2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJWHE3DCNRVGY> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[lrnv]

Now the tests are passing, I will merge this if @Santymax98 you agree ?

[Santymax98]

Yes, I agree. Thank you for your help El El lun, 20 nov 2023 a la(s) 12:27, Oskar Laverny < ***@***.***> escribió:

…

Now the tests are passing, I will merge this if @Santymax98 <https://github.com/Santymax98> you agree ? — Reply to this email directly, view it on GitHub <#73 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A7PTWNQQFTA6OPA5OJ6Q6A3YFNZG5AVCNFSM6AAAAAA7P6IUQ2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJZGI4DGNJXGU> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[lrnv]

My pleasure ! We fixed a lot of stuff with these tests that is really good.